September 2019 | ISE Magazine 33
Reducing surgical residents burnout
using neurofeedback
Study measures results of medical workers’
depression before and after treatment
By Lukasz M. Mazur, Alana Campbell, Prithima Mosaly, Karthik Adapa, Dr. Ian Kratzke,
Dr. Lawrence M. Marks, Dr. Samantha Meltzer-Brody and Dr. Timothy M. Farrell
34 ISE Magazine |
Reducing surgical residents’ burnout using neurofeedback
In recent years, changes in the healthcare industry have
increased scrutiny on financial productivity, quality of
care, patient safety and care outcomes taking away au-
tonomy from clinicians. It is no surprise that national
studies suggest that burnout and depression rates among
surgeons range from 30% to 38% and have increased
over the past five years to more than 50% (“Multiple Insti-
tution Comparison of Resident and Faculty Perceptions of
Burnout and Depression During Surgical Training,” Michael
L. Williford, Sara Scarlet, Michael O. Meyers, et al, JAMA
Surgery, 2018).
Burnout is a stress-related syndrome and is characterized
by emotional exhaustion, depersonalization and a decreased
sense of personal accomplishment. The prevalence of burn-
out and depression are greater among residents than among
medical students, physicians or college graduates of similar
age. Furthermore, surgical residents with high burnout and
depression are at an increased risk for suicidal ideation.
Training of surgical residents involves complex and de-
manding cognitive activities including multitasking, clinical
reasoning, problem-solving and overall information process-
ing, all of which results in high cognitive workload. Studies
suggest that cognitive workload is impaired in burnout and/
or depressed residents exposed to high task demands. There-
fore, surgical residents with burnout or depression are more
likely to commit medical errors that can lead to patient safety
issues, including patient harm.
To optimize cognitive workload, neurofeedback proto-
cols are being increasingly used in diverse fields, including
healthcare. Neurofeedback is a scientifically based technique
that allows the brain to train its self-regulation skills. The
process is based on operant conditioning and it is often de-
scribed as “exercise for the brain” that increases the efficien-
cies of specific brain functions and enhances cognitive skills.
To date, no previous work has investigated the efcacy
of neurofeedback protocols in improving cognitive work-
load, performance and symptoms of burnout and depression
in surgical residents. We herein present the results of an in-
novative pilot study intended to assess the impact of neuro-
feedback on the cognitive workload of surgery residents with
burnout and depression. Notable improvements in cognitive
workload and growth areas following the neurofeedback
treatment were recorded, suggesting a possible return to less
burnout condition.
Methods used in the study
From June to August 2018, 15 surgical residents with burn-
out – a Maslach Burnout Inventory (MBI) score of more
than 27 – and depression – a Patient Health Questionnaire-9
Depression Screen (PHQ-9) score of more than 10 – from
one academic institution were enrolled and participated in
this institutional review board (IRB) approved prospective
study. Ten residents with more severe burnout and depres-
sion scores were assigned to a neurofeedback treatment, and
five others were treated as controls.
Each participant’s cognitive workload (or mental effort)
was assessed initially, and again at an eight-week interval, via
electroencephalogram (EEG) with the oscillatory power re-
corded while the subjects performed a computerized n-back
working memory task. This task, a widely used measure for
the assessment of working memory function, involved indi-
cating when a current stimulus (a picture) matched the one
from n steps earlier in a sequence (e.g., “1-n” requires that
participants had to remember the picture presented one im-
age previously, and so on). It used E-Prime software, with
an inter-stimulus interval of 1,500 milliseconds and stimulus
presentation time of 500 ms, while seated in front of a com-
puter in a sound and light attenuated room, at 72 degrees
The treatment consisted of eight validated alpha-theta
neurofeedback sessions, each 35 minutes long, during the
eight-week interval. The alpha-theta protocol was divided
into two separate periods: 1) Pz alpha/theta training period
(eyes closed, deep relaxation for 24 minutes) with the inhibit
frequencies set to 2-4 hertz (Hz) and 15-30 Hz and the re-
ward frequencies set to 5-7 Hz and 8-11 Hz; and 2) C5 beta
awakening/arousal period to ensure proper mental activa-
tion before releasing participating residents back to the clini-
cal work areas (eyes open for nine minutes) with the inhibit
frequencies for beta C5 training set at 1-12 Hz and 20-30
Hz and the reward frequencies set to 15-18 Hz. Throughout
the neurofeedback periods, the impedance was maintained
below 10 kiloohm.
Overall, each session began with instructions for the par-
ticipants to remain relaxed and still for approximately 20 to
40 seconds as BrainPaint neurofeedback software gathered
baseline measures for the reward and inhibit frequencies.
Next, participants were instructed to perform their “best
to maintain their cognitive state for deep relaxation dur-
ing alpha/theta protocol as guided by the respective reward
frequencies. No specific instructions were provided for the
awaking/arousal period.
During both periods, when rapid increases in the 1-12 Hz
and 22-30 Hz frequency ranges were 30% greater than the
amplitudes recorded during the baseline period, participants
were notified by the software via verbal and visual feedback
about potential excessive movement or muscle tension. EEG
recordings were made on a BrainVision Recorder, with
BrainAmp 32 channel system sampled at 500 Hz filtered
online between 0.16 and 100 Hz, then were preprocessed
and analyzed in EEGLAB and with Matlab scripts. All EEG
data were pre-processed and analyzed by an expert cognitive
scientist with specific expertise in EEG data processing using
custom EEGLAB and Matlab scripts. Data were downsam-
September 2019 | ISE Magazine 35
pled to 256 Hz and independent com-
ponent analysis (ICA) was run to help
identify and remove artifacts (e.g., eye
blinks) and segmented around stimuli
by type from -4 to 4s. Analysis of vari-
ance (ANOVA) with parametric boot-
strapping tested for signicance between
the degree of change (pre- vs. post-) in
the treatment and control groups. All
P values were two-sided, with P < .05
deemed significant.
In addition, to proactively mitigate
study risks, we asked each participant to
report daily on their overall status (e.g.,
amount and quality of sleep, feeling of
being depressed, thoughts of suicide).
Participants used a four-point scale
ranging in general from better than yes-
terday, same as yesterday, average, worse
than yesterday, to one of the worst days
ever. Participants were also asked to
identify three to four specic growth ar-
eas in which they would like to improve
over the eight-week period (e.g., con-
centration on task, stress levels, temper,
overall energy), using the learnings from
the neurofeedback sessions.
Thus, starting with second session,
participants reported progress using a
-100 to +100 absolute scale for each
selected growth area. For each session,
we averaged these absolute differences
in scores to generate a composite score
reflecting the overall perceived progress
by the intervention group on their indi-
vidualized growth areas. With this data,
we analyzed for the correlation between
the time (sessions) and average absolute
improvements in growth areas as report-
ed by the participants.
Results from neurofeedback sessions
Figure 1 demonstrates overall EEG power per time (x-axis)
and frequency (y-axis) for the n-back target stimuli during
the pre- and post-assessments in the control and treatment
groups. Both groups show relatively high cognitive workload
in the pre-assessment, with somewhat inefficient theta (4-8
Hz) and alpha (8-11 Hz) activity (represented by more dark
blue color).
After the neurofeedback intervention, the treatment group
showed a significant (p < 0.01) improvement in cognitive
workload during the working memory task with changes
in EEG oscillatory theta and alpha power. These differences
were not noted in the control group.
Throughout the study, we experienced only two instances
when two different participants reported concerning trends
in their feeling of depression and were immediately contacted
by an experienced psychiatrist to intervene as needed. Fortu-
nately, no major interventions were needed, and in both cases
participants were allowed to continue with the study proto-
col. Participants also reported significant improvements in the
growth areas – significant correlation between time (sessions)
and absolution improvements in growth areas (p < 0.01).
Neurofeedback, before and after
Overall EEG power per time (x-axis) and frequency (y-axis) for the n-back target stimuli
during the pre- and post-assessments in the control and treatment groups.
36 ISE Magazine |
Reducing surgical residents’ burnout using neurofeedback
In this study, there was a notable change in cognitive
workload following the neurofeedback treatment, suggest-
ing a return to a more efficient neural network. To the best of
our knowledge, this is the first study in surgical residents to
demonstrate improvements in cognitive workload as quanti-
fied via brain EEG patterns following a neurofeedback treat-
In all 15 subjects, initial baseline activity exhibited rela-
tively inefficient cognitive workload, especially as repre-
sented by theta (4-8 Hz) and alpha (8-11 Hz) activity. Such
a pattern, while hypothetical in residents with burnout, is
worrisome and could reflect the need to recruit additional
cognitive resources to complete the working memory task.
For example, quantitative EEG studies, while not focus-
ing on cognitive workload, suggest that inefficient theta (4-8
Hz) and alpha (8-11 Hz) activity is associated with post-
traumatic stress disorder and elevated theta is associated with
impaired working memory performance in patients with
PTSD. Empirical evidence from prior randomized control
trials and multiple pilot/exploratory studies demonstrates the
effectiveness of neurofeedback in the treatment of PTSD –
e.g., significant symptom improvement compared with con-
trols; 70% of participants in the treatment group not meeting
the diagnostic criteria for PTSD following the neurofeed-
back treatment (“EEG-Neurofeedback as a Tool to Modu-
late Cognition and Behavior: A Review Tutorial,Frontiers
in Human Neuroscience, Stefanie Enriquez-Geppert, Rene
J. Huster, Christoph S. Herrmann, 2017). There were also
notable improvements in growth areas, suggesting tangible
benefits to participants.
For example, one subject reported improvement in qual-
ity of time with family and children: “I noticed that on the
days with neurofeedback secession to have better interaction
with my family. For example, last week I was calmer with
my children and quality time was much better. My short fuse
was longer.
Another participant reported control over self-regulation
of negative thoughts: “Neurofeedback made me more con-
scious of my internal feelings and provided me with more
self-control. This allows me to put my head in the right
space. Also, it helps me relax faster.
However, not all participants directly benefited from this
study. One participant reported the following: “Neurofeed-
back was a good opportunity to be mindful. Disconnecting
and meditative aspect was helpful. However, I noted no spe-
cific changes in my selected growth areas. I still have trouble
focusing on the task, my mind wonders a lot and I have a
hard time falling asleep.
This exploratory study has several limitations. First, the
results are based on one experiment with few subjects and
without a “sham” neurofeedback control group. Given the
extensive training needed in the study (eight neurofeedback
The 6 types of brain waves
Brain wave speed is measured in hertz (cycles per second) and
divided into bands delineating slow, moderate and fast waves.
Here are descriptions for the different types of brain waves:
Infra-low (below 0.5 Hz). Infra-low brain waves, also
known as slow cortical potentials, are thought to be the basic
cortical rhythms that underlie higher brain functions. Their slow
nature makes them difficult to detect and accurately measure.
Delta (0.5 to 3 Hz). Deltas are slow, loud brain waves,
low frequency and deeply penetrating like a drum beat. They
are generated in deepest meditation and dreamless sleep. Delta
waves suspend external awareness and are the source of empathy.
Healing and regeneration are stimulated in this state.
Theta (3 to 8 Hz). Thetas occur in sleep and are also
dominant in deep meditation. They are the gateway to learning,
memory and intuition. It is that twilight state experienced as we
wake or drift off to sleep.
Alpha (8 to 12 Hz). Alpha brainwaves are dominant during
quietly flowing thoughts, and in some meditative states. Alpha is
the resting state for the brain and aids overall mental coordination,
calmness, alertness, mind/body integration and learning.
Beta (12 to 38 Hz). Beta brainwaves are present in our
normal waking state of consciousness when attention is directed
toward cognitive tasks and the outside world. Beta is a present
when we are alert, attentive, engaged in problem-solving,
judgment, decision-making or focused mental activity. Beta waves
are further divided into three bands: Lo-Beta or Beta 1 (12-15 Hz),
a “fast idle;” Beta 2 (15-22 Hz) high engagement; and Hi-Beta or
Beta3 (22-38 Hz) highly complex thought, anxiety or excitement.
Gamma waves (38 to 42 Hz). The fastest of brain waves
relate to simultaneous processing of information from different
brain areas. Gamma waves pass information rapidly and quietly.
September 2019 | ISE Magazine 37
sessions, 35 minutes each in length over an eight-week pe-
riod), only a modest number of participants could receive the
treatment. Such is often common in these types of studies.
Second, nonrandom allocation of subjects led to an inten-
tional higher level of initial burnout-depression in the treat-
ment group, thus raising the possibility that they had a larger
potential to improve.
Nevertheless, the significant improvement in cognitive
workload following the neurofeedback treatment suggests
that this innovative approach warrants further evaluation as
a potential intervention to address burnout-depression con-
cerns for surgery residents.
Future studies could also examine the degree to which
burnout and depression symptoms in surgical residents cor-
relate with specific alterations in EEG or other neural activa-
tion patterns, as well as behavior outcomes.
This study was supported by funding from the University of North
Carolina Health Care System. The authors thank David Planting
for his assistance in administering neurofeedback sessions and research
participants for their time and effort.
Lukasz M. Mazur, Ph.D., is an associate professor and a director
of the Healthcare Engineering Division at the Radiation Oncology
Department in the UNC School of Medicine at Chapel Hill, North
Carolina. He earned his bachelors degree, masters degree and Ph.D.
in industrial and management engineering from Montana State Uni-
versity. His research interests include engineering management as it
pertains to continuous quality and patient safety efforts in healthcare
and human factors engineering with a focus on workload and indi-
vidual performance during human-machine interactions.
Alana Campbell, Ph.D., is an assistant professor in the Department
of Psychiatry at the University of North Carolina at Chapel Hill,
North Carolina. She earned her bachelors degree in human develop-
ment and family studies from Cornell University, a masters degree in
psychology from Colorado State University and a Ph.D. in psychol-
ogy and cognitive neuroscience from Colorado State University.
Prithima Mosaly, Ph.D., is a research assistant professor in the De-
partment of Radiation and Oncology at the University of North
Carolina at Chapel Hill, North Carolina. She earned her masters
degree in health policy and management from UNC-Chapel Hill
and Ph.D. in industrial and manufacturing engineering at the Uni-
versity of Wisconsin Milwaukee and is a certified Human Factors
and Ergonomics Professional.
Karthik Adapa, M.D., is a graduate research assistant for the Caro-
lina Health Informatics Program at the University of North Carolina
at Chapel Hill, North Carolina.
Ian Kratzke, M.D., practices in the Department of Surgery at the
University of North Carolina, Chapel Hill, North Carolina. He
earned his M.D. from the Medical School: State University of New
Lawrence M. Marks, M.D., is a radiation oncologist at the Univer-
sity of North Carolina, Chapel Hill, North Carolina. He received
his medical degree from University of Rochester School of Medicine
and Dentistry and has been in practice for more than 20 years. He is
chair and Dr. Sidney K. Simon Distinguished Professor of Oncol-
ogy Research.
Samantha Meltzer-Brody, M.D., is an associate professor at the
University of North Carolina, Chapel Hill, North Carolina. She is
the Ray M. Hayworth and Family Distinguished Professor of Mood
and Anxiety Disorders; executive medical director of the UNC Well-
Being Program; and director of the Perinatal Psychiatry Program,
UNC Center for Women’s Mood Disorders. She earned her bach-
elors degree in biology and psychology from Simmons College and
her M.D. from Northwestern University Medical School.
Timothy M. Farrell, M.D., is professor of surgery and director of
minimally invasive surgery at the University of North Carolina,
Chapel Hill, North Carolina. He earned his M.D. from the Uni-
versity of Medicine and Dentistry of New Jersey.